135
Cognitive Processing in Bird Song Development and Use Psych 1090 Lecture 10

Cognitive Processing in Bird Song Development and Use Psych 1090 Lecture 10

  • View
    217

  • Download
    0

Embed Size (px)

Citation preview

Cognitive Processing in Bird Song Development

and Use

Cognitive Processing in Bird Song Development

and Use

Psych 1090

Lecture 10

Oscine birds learn their songs, just like human children learn their

language…..

Clearly, there are innate predispositions to learn both song and

languageand particular brain areas are involved

in the processes for both birds and humans

But learning does occur…..and usually involves some cognition

And, as we’ll see, it isn’t only what the bird learns that is

important

but also the way it chooses to use its vocalizations

the choice of which song to sing or the way in which it is sung can

have significant effects on other birds

But, with respect to aquisition, birds need to learn

from one’s neighbors

what constitutes the communication code

which notes or songs to sing

what order in which to sing them

from whom to learn the code

from one’s father

from a combination of tutors

And, it appears that all these behavior patterns involve, to some

degree, some forms of cognitive processing

And how to use the communication code

to recognize individuals

to attract mates

to defend territories

Specifically, the natural world is an extremely complicated place

of myriad interactions--

some obvious, some hidden--

but all of critical importance if one is to understand its

workings

Information must be processed, sorted, ignored or acted upon by

all creatures

And, as we’ve said before, fully programmed responses would be almost useless in dealing with a

world of myriad situations

Flexibility in processing in crucial for success

John Smith (1998) sums it up:

“…an animal has become flexible when it can process individually acquired knowledge, seek further information and hold a range of options open as it works to anticipate the course of an event. Such an individual can fine-tune and modify its responses. It can evaluate unexpected information and cope better with variable environments than can the ‘releaser-driven’ individuals posited by early ethologists.”

We’ll start with acquisition and then go on to use

But remember that these are going to be inter-related issues

for some avian specis

Looking at birds’ choice of from whom they will learn their

songs….

doesn’t necessarily reflect cognitive any levels of

processing…

Many times birds just learn to sing from their fathers…

But then they do differentiate their song enough to allow themselves

to be recognized individually

Simply because they hear their father’s song more frequently than any other song around…

So that they don’t sound like dad or their brothers

But in other cases, birds choose to learn from territorial neighbors

particularly ones that appear most dominant in the area

In such cases, cognitive processing is clear because they

evaluate the various singers

But even then the issue isn’t entirely settled

because there are factors involving whether the birds

stay where they learned their song

or disperse to another area with a different song tradition

And although “tradition” generally refers to a song

dialect

that is, a variation on the species-specific format

it can also sometimes mean that the newcomer has to fit into dominance system in a given

area

All birds do learn to something about song in

their first year

And some actually start to sing in that hatchling year

Whereas most won’t actually come into full song until they are

almost a year old…their next spring

In general, the pattern of song learning proceeds as was described by Marler and his colleagues in the

1960s:

The differences are in the details…

Different species learn at somewhat different rates, different numbers of songs, and with more

or less flexibility

but the basic pattern is the same

Birds like indigo buntings put down song memories in their first year,

but then migrate away and return to areas that are not necessarily those in

which they were hatched

they then switch their songs to replicate that of the most dominant

male in the new area

Birds like chaffinches, however, try to sound different from their

neighbors….And appear to be able to learn songs in both their first and second years….

According to recent work, they sound like neighbors who are 3 or 4

territories away….

Given that actively learning songs at such a distance is unlikely,

What probably is happening is that birds sort each other out

Now, given that some territories are better than others, one could begin to

imagine the types of dominance games that must be played in such a

situation!

and choose to nest away from neighbors that sound like they

do….

Which, if nothing else, means that both species, in different ways,

The process involves the transitive inference work we’ll discuss next

week

and suggests at least some level of cognitive processing

have to evaluate the interactions among at least some different males

in the area

We’ll talk more about such aspects in a bit…

But you also need to know something about the flexibility in the song acquisition process that

was not initially expected…

Birds were once thought to have a critical period

and also a restrictive song template

The idea was that the birds heard lots of song when they

were first hatched

because the adult birds in the area were still defending their territories

from other birds

The idea of a template was to filter out all the allospecific songs so that they wouldn’t

learn the ‘wrong’ song

And the idea of a critical period was that they’d be sure to get

song when dad was singing close to the nest, again insuring the

‘correct’ song

The template and critical period ideas came from studies done on white-crowned sparrows raised in

social isolation and given tutor audiotapes

so that researchers could control exactly what the birds heard and

when they heard it

Under such conditions, white-crowned learned best between

days 10 and 50

and ignored songs of other species that they heard on tape

But, of course, the conditions did not replicate reality

It turns out that white-crowns are much more flexible when their tutors are live:

If they hear only one tutor song to day 50, they sing that song…..

But if they have social interactions with adult birds at least up to day

100,

They will switch and sing more of the song they hear later

One can argue that what is occurring is simply an ‘overwriting’ of the

earlier song by more recent material

Or that learners are just ‘remembering’ song elements of the later tutors from some innate

storage bin….

But those explanations are not entirely enough….

First, birds that do not hear any tutor sing a totally bizarre song,

That is, some memories have to be laid down before they can be

recalled

so they can’t just create songs from innately stored material

Second, if white-crowned sparrows are put in situations with a bird that sings the original song it learned….

it will switch back to that song; that is, revert to the song it would seem to

have forgotten…..

and can actually switch back and forth between different songs depending on

what it hears

Now, one might conclude that the bird is just reacting in a species-

specific manner, replicating, tape-recorder like, what it has just

heard….But, given that white-crowned

sparrows normally sing only a single song, such behavior is not likely to

have been hard-wired in

But rather the bird is processing what it has heard and reacting

appropriately….

Other birds give clearer evidence of processing in terms of how they learn and sing their

songsparticularly when they have a very

large repertoire

large numbers of songs are both difficult to learn and difficult to

maintain in a given order so as to retain immediate variety (not repeating what was just sung)

Some of the best studied birds in this regard are nightingales

they learn strings of over 60 different songs

after singing each song, the bird is making a decision about the next song

in the sequence—

and the sequence of songs in a their song bout actually represents a

sequence of behavioral decisions

that is, the birds engages in some form of cognitive processing—and humans can abstract rules about

these decisions

Moreover, careful study has shown that the ordering of the songs is of biological importance to the singer

and not just an artifact of the analysis done by a researcher

When singing, the nightingale actually chunks its repertoire into groups (or, as the researchers say,

“packages”) of about 3-5 songs

The birds maintain the serial order of the packages but not necessarily that

of songs within each package

So researchers see a pattern like the following in terms of learning and

recall….

1-5

6-9

10-16

17-20

21-27 28-

3435-40

41-46

47-54

55-60

61-63

64-70

17

2019

18

Rendition #1

17

2019

18

Rendition #2

Further studies show conclusively that storage and production process does not simply involve paired associations

between song-types

Otherwise, birds could not go from, for example, no. 19 in one packet to no. 22 in another packet at one time, and a different between-packet order at

another time

Although Todt and Hultsch do not suggest that this intra-package ordering could be important in song-matching

which we’ll see is a critical issue for other birds that we’ll discuss

in this lecture

the possibility exists

But how do the birds learn these long lists—particularly if they hear lots of birds singing around them?

But what happens if you give a bird a long list—say, of 60 songs

or several such lists in which songs closest to one another do

not have any particular similarity…

And what if the songs were not separated in any special ways….

how would a nightingale form a repertoire of its usual 100 or 200

songs?

To figure it out, you do a series of clever experiments like those of

Todt and Hultsch

They began by exposing young nightingales to several different

long series of songs in the laboratory, called master strings

By permuting song orders in the different strings, the number of

song repetitions, and other variable such as social interaction,

they determined what was needed for the birds to acquire

the songs

and how birds ordered these songs in their singing bouts

Specifically, as would humans, the birds classified the large numbers of songs heard in master strings into smaller, more manageable

chunks.

The size of these chunks or “packages” , of three to five

songs,

and is probably constrained by short-term memory

relates directly to the chunks seen in singing

The songs in each package seem to be learned at about the

same time

and to arise early in the process of song development

and, as in the final version, song order is quite variable within the

packages

But even as the birds learn, the packages are themselves arranged

in a stable order,

and this order reflects the context, or the master string, from which the

songs were learned

This package ordering, however, in contrast to the structure of the packages themselves, emerges

relatively late in the song development process

There are no limits to the number of packages that are strung

together,

and the number is determined by the length of the master string

that the bird hears

But that’s not all: often, late in development, a nightingale invents a new set of songs…

but the set derives from those already in a given package

Interestingly, birds that hear the same master strings may

have different package boundaries

and thus somewhat different overall song order, maybe

for individual identity

but the rigorous general form of the packages will be

maintained

that is, packages never vary from the 3-5 song limits

Now, sometimes birds were given multiple master strings…

when such birds are practicing during the acquisition process,

the intervals between imitations acquired from the same master

string or context group were significantly shorter

than the intervals between imitations acquired from different

context groups.

That’s important, because it suggests how the birds store the

strings….

That is, if the time needed to access the stored

representation of song-types varies

depending on whether retrieval is from within a given context group or from a different context group…

Then there is data providing evidence for hierarchical memory and likely cognitive processing

In other words, the birds appear to be categorizing the songs with what would likely in the wild be

songs heard from a particular individual, or at a particular time or

location

Such data couldn’t be used to argue for episodic memory

directly,

but suggests the possibility…

Obviously, in the wild, they hear strings not only from their father, but also from other neighboring

males.

One possible problem was that if, in the lab, the master strings are

separated by less than five minutes and repeated only 20

times, the birds can’t remember the

context….

In the wild, the time separation is likely to be as short as 5 min…but the repetition rate is likely greater

But the number of exposures does seems to have an important effect

on song learning

Although birds can learn some songs after hearing them as few

as 5 times,

they seem to need to hear them at least 15-20 times to reproduce

them clearly

Which makes ecological sense…

Why bother to learn the song of a bird just passing through, or that

may have died?

A bird wants to learn the best set of songs from the best songster

It also seems that twenty exposures are necessary in order to get the song order down pat…

If, however, a master string is presented for something like 100

repetitions,

the package effect (variability within packages) is smoothed out

in the sense that the order of songs within the packages

becomes much more defined

and thus the overall order is more stable and the package boundaries become less

important

which is also important….

Because it suggest that birds are, indeed, using packages as an aid

to memory…

When exposures are limited, and a bird hasn’t been given much

chance to form the serial order,

it chunks information into packages much as do humans who

are faced with processing long strings of information in a limited

time

Furthermore, if the song interval within the master string is

doubled,

from a normal rate of 4 seconds to about 10,

that is, when information is given with additional time for acquisition (and likely

processing)

the birds have less need to chunk

This result again suggests that package formation, or ‘chunking’

is a cognitive process found in nightingales

And other data show even more evidence for cognitive

processing

Data obtained during the song development process suggests that

the nightingales are retrieving information from memory

and are retrieving this information in a manner

consistent with a specific pattern of storage

Early in development, nightingales produce material

that is either

easily identifiable as something they heard in a master string (identifiable patterns, IPAs)

or not (unidentifiable patterns, UPAs), which might be total

inventions

As they get older, IPAs increase and UPAs decrease….

but the birds are not randomly producing this “new” IPA

material The IPAs are temporally clumped, and new material is embedded in

existing IPAs

rather than being derived from UPAs

The pattern of storage does not, however, reflect the order

of master string input …

Material from strings heard early does not appear before

material from strings heard later

So, what does this all mean? How can we describe the

process?

Hultsch and Todt (1989c) suggest a three-part system:

short-term memory

recognition memory

submemories

Short-term memory is constrained

By both the number of items to be remembered and the time allowed for retrieving them

Thus, to retain songs from long master strings (that are presented

only ~20 times)

in short-term memory prior to transfer to long term memory…

the nightingale, like humans, segments the serially coherent

master strings into different packages of information

that is, into collections of 3-5 units

Then recognition memory identifies the songs of a package

as eithernovel (new to short-term memory)

or familiar (already in short-term memory)

if the former, the material is transferred to a submemory

The submemory processes the package into a formal unit that can then be recognized by short-term

memoryThus, for each novel master

string,a group of submemories is

formed

But what if the package is familiar?

Then recognition memory transfers the material into an existing

submemory, which is thereby strengthened

And, because long master strings are learned as effectively as short ones, even when heard only ~ 20

times processing of the group of

submemories likely occurs in parallel

However, because each package is connected to a particular context (i.e., master string)

an additional process has to be postulated

that somehow associates and connects the sets of submemories that were developed from a given

master string

As it turns out, the first exposure to each master string plays a key

role in acquisition

Birds remained fairly still and attentive during their first

exposure to a master string

and moved around more as the number of string presentations

increased

but then they quieted down when a new set of song-types—a

new master string—was presented

Habituation/dishabituation effects

Biologically, this scenario makes sense….

Pay attention to new stuff, even just new ordering of old stuff

And, of course, I haven’t even touched on the structure of the

individual songs…

It’s clear that these are incredibly complex learning issues for these

birds

Now, all this ordering is important for song retrieval as well…

In nature, birds have about one second to identify a neighbor's

song and

at the same time also select and retrieve a song of the same type

from his own repertoire

Even tho’ nightingales don’t match songs as much as other birds we’ll

discuss

They still need to be able to access what they have, and all this chunking and packaging

helpsBasically, their memory

processes seem quite like those of humans learning long

lists

And this memory form is important for other birds as

well…

List learning also appears quite important in marsh wrens….

These birds can sing from 100-400 songs in a specific order;

they share many of these songs with their neighbors

who sing them in different orders

Now, when researchers first started studying songbirds,

they assumed that large repertoires were used only to

prevent habituation

So that birds didn’t just tune one another out…and that might be

partially true for nightingales

But, as we shall see for marsh wrens (and later for song

sparrows)

such is not the case…

A lot depends on the particular song that is sung and thus

deliberate choice is important

So, each wren has to know the order of its own ~200 songs and the order of 3-4 neighbors’ ~200

songs….

Now, marsh wrens nest in fairly small territories, which means

they have 3-4 neighbors…

And they match songs to defend these territories…

In territorial disagreements, they perform a sort of marsh-wren

group poker….

Bird A will match the song of neighbor B and then sing the

next song not in his, but in B’s list….

tending to go through at least portions of their repertoire in lock-

step

Sometimes all the males in close proximity joined in, all using the

same song types

Given that each male, when singing alone, generally has a somewhat different sequence from other birds in the area,

the wrens seem to be making active choices about what to

sing based on their processing of information from the other

birds around them

they also have to recognize, if not sing,

other songs of their neighbors that are not in their own

repertoire

in order to help them recognize the particular neighbor that is

singing

So that is a huge memory and processing load

When the researchers examined the singing patterns,

They initially did so by testing only two birds in the laboratory

training them on different sequences of similar, not

identical, strings

B can chose to match this song and then continue, or just

continue

So, when B heard A sing one of his (B’s) song,

which shows that he is subordinate to A

Or B can sing the song in A’s repertoire that follows this song,

So you hear something like this…..

which shows that he is challenging A for dominance

Bird A Bird B

10

13

12

11

27

14

28

29

30

31

Bird A hears B’s 27, which is his 11; sings B’s 28, which may be his 14

B can sing his own 29, or challenge A by singing A’s 12, which could be B’s 2 or his 37…

?

And this occurs for all neighbors in the area….3 or 4

birds, generally

And this can be going on not sequentially but simultaneously

think of the speed at which these interaction have to occur and decisions have to be made

Marsh wren song rates are about 20 songs/min

up to four other males are signing at roughly the same time

So, like the nightingale, the wren may be hearing another

song every second

has to figure out what to sing in reply to that song in that 1 sec

timeframe

Possibly, too, birds in neighboring territories are eavesdropping on these

interactionsusing the information obtained to determine the relative worth of

their neighbors

Now, for the chestnut sided warbler, we go to

SciAm…

Sometimes songs are altered based on feedback from the

female, as in cowbirds

Other types of input are used for choice of song…

Now, cowbirds are nest parasites, so the males have to learn their song from someone

other than dad…

So that, unlike other birds we discussed, they hear only the

‘wrong song’ in their first summer….

Thus they must choose appropriate flocks for the fall and winter

…without necessarily having seen other of their species….

They get the basics from hanging out in juvenile groups

after fledging

Then they try out their song on the various females that they

come across

And they better get the local song right…

If the male sings a song from the wrong tradition,

he’s toast

What would be of interest would be experiments to see if the

male would choose a cohort in a “bilingual” environment…

But not if another, more dominant male, is in the area

But, even in the tradition, the male will sing a number of different

songs, wait for the female to give a wing-

flip to show she likes that one

and the male will then use the preferred song almost exclusively

In an aviary, juvenile males housed with adults tended to associate primarily with other

juveniles,

But sang to adult males and adult females…

They then exhibited normal courtship and aggressive behavior

patterns

In contrast, juveniles that were put into a situation in which they did not have any interactions with

adultsbasically didn’t know to

approach females or avoid adult males in the breeding season….

Suggesting all such behavior was learned…

And, unlike juveniles w/ adult interactions,

They didn’t know what to do with their songs!

their song crystallized sooner, and were more potent….

which isn’t necessarily good when they are then put into a normal flock

They could get clobbered by the dominant males

And if adult males were housed without juveniles, who would

challenge them,

They later did not compete well amongst each other for females…

i.e.,they seemed to lack the ability to evaluate interactions—unable to perform appropriate cognitive

processing

It would seem that some level of processing exists, even if on an

unconscious level

Other experiments show how these different patterns can be affected by

change, suggesting that

The behavior develops in an ‘on-line’ manner, depending upon what is

currently happening….

Birds also appear to use some form of cognitive processing

when recognizing their neighbors…ie., same/different

Remember how tough it was to demonstrate true same/different

behavior…

Rather than simple matching behavior or other simpler actions

For song birds, natural behavior patterns of individual recognition, vocal dueling and song matching require same/difference-based

discrimination

implying that such discrimination is an adaptive trait

In the laboratory, however, behavior is not easy to demonstrate….

For example,

budgerigars’ apparent discrimination of similarities and differences in canaries’ calls may be based instead on learning the unique characteristics of individual calls similar results were found in song discrimination studies on cliff and barn swallows and great tits

That is, match-to or different-from-sample

Song sparrows appeared to make fine acoustic distinctions among conspecific

song syllables in a habituation/dishabituation experiment

but data must be interpreted carefully with respect to same/different because

these birds can memorize large numbers of song types

That is, again work on the basis of MTS or NMTS…

Studies showing how different sound features differentially affect conspecific song recognition and discrimination of conspecific from allospecific song for

each species

demonstrate just how difficult it is to design experiments to show that birds

understand same/different the way Premack would require

e.g., intro notes versus final trills

Field sparrows differentially weigh information in various song features

and thus results could depend on which feature an experimenter

chooses

Specifically, experimental design may not demonstrate a subject’s

perception of same/different

But many warblers can tell not just whether a song is from a

familiar neighbor or an unfamiliar stranger

That is, MTS…

but also whether the song from a familiar bird is coming from the

wrong territory….

And remember this information from one year to the next!

which also implies some form of spatial encoding and cognitive

processing….

At least a very good memory…

Song sparrows in the wild have a much smaller repertoire than

nightingales and marsh wrens…5-10 songs

So, to understand these birds, you need to know a bit about their

ecology

and they engage in a complex form of individual recognition…

Initially, researchers didn’t think song sparrows actually did have

individual recognition…

or even neighbor-stranger recognition

Because they would react like banshees to anything played to

them in some experiments…

And did show a neighbor-stranger difference in other

experiments..

Stoddard, in an elegant set of experiments, finally learned that the problem was that these birds had exquisite neighbor-stranger

and individual recognition

And that they were extremely sensitive to their territorial

boundaries….

So that if the experimenters played anything within their territory…

neighbor or stranger…

They got the banshee response

But IF they played the neighbor song outside the boundary,

they got a mild response

And a stronger response to the stranger…Why?

Song sparrow populations have a lot of birds called ‘floaters’…

Birds that are somewhat familiar, but that are without territories

And these floaters are always on the lookout for a territory that isn’t

well defended

So any intrusion, even by a somewhat familiar bird, is seen

as a serious threat

Now, for song sparrows, in many populations, unlike marsh wrens,

neighbors do not share song types, and therefore cannot type-match….

by using a song from his repertoire that was similar in some way the

birds recognized

But a bird lacking a true type match could still song-match a stimulus

song

So, if a bird hears a song that is similar to one or more of his songs, but an exact match to

none of them, it replies at rates significantly

above chance with whatever in its own song is the closest match…

e.g., a song with a ‘double buzz’ intro, or one with a specific type

of trill….

Again, suggesting some level of processing…

the bird cannot reflexively (or mindlessly) sing back exactly what

it heard

but somehow rather very quickly must search through its repertoire

(those 5-10 songs) to find the closest match.

Even if one argues that the bird is not basing its choice on a

complex level of same/different discrimination

but ‘merely’ on the basis of familiarity….

the task is not simple:

particularly in early spring, the territory owner can be faced with a large number of such intruders…

each of which has its own repertoire

and the owner must quickly choose the appropriate reply or

stand to lose its territory

Moreover, the intruder might try more of its repertoire in an

attempt to outsing the owner

Think of the interaction as a kind of poker game, in which you can’t

be sure of the cards that your opponents are holding….

and you begin to understand

But, of course, as the season progresses, the birds start to learn about their neighbors

And they couldn’t breed or feed young if all they did was defend

their territory

So they eventually engage in a kind of détente….

So, as they season progresses, the shift from song-type matching to repertoire matching of neighbors

In essence saying “OK, by singing something I know you know, I

don’t really need to fight you, just remind you that I am still me and

keep out”

A kind of “dear enemy” situation

Remember, the neighbor can still be a threat, and will invade

if he hears no response

So they are carefully grading their responses, carefully choosing

their songs after learning what their neighbor sings…

And they still patrol their border w/ unshared songs

To let birds in distant territories or floaters know that someone is

there w/o annoying their neighbor

But each bird has to remember the level of aggression of each

neighbor

as well as the several songs of each of its neighbors (3 or 4 of

them)

and must always be vigilant against floaters

whose songs may be somewhat familiar

Remember, the détente is based on the intimate knowledge of what

is normal and expected in the situation

and knowing when the norm is violated

None of this can easily be explained as a simple stimulus-

response

But involves a lot of information processing

And shows complexity even with a relatively small repertoire

Which is why scientists can fake them with playbacks…

Of course, the song sparrow cannot tell what bird is singing, only

something about the song that is being sung

Thus the sparrow isn’t recognizing a particular neighbor, just what

should or should not be heard from a specific location

If they hear a playback in their territory from a stranger that uses a

song very similar to that sung by their neighbor

They still respond as though to a stranger…..

Thus a floater attempting some for of deceptive mimicry would be

thwarted….Suggesting that such information

processing is evolutionarily adaptive

Studies on chickadees (a North American bird like the great tit) and

zebra finches also show that the abilities involved in individual

recognition are learned….Birds that are raised as isolates, when compared with birds that have been captured in the field take much more time to learn discriminations in the

lab….And you would think that lab-raised birds would have the advantage of familiarity with the environment….

Because great tits, like the buntings, learn their songs and adjust their songs when they arrive on their

breeding grounds,

Great tits present another interesting example of some level of

processing….Researchers found that they shifted

the frequencies of their songs in noisy urban environments

Results therefore suggest that birds learn to use a restricted range of their

spectral capacity in response to frequency-dependent interference

from local noise conditions--adjusting song to territory instead of territory to

song.

It wasn’t just a matter of genetically pre-determined singers being

selected…

Of course, at some point, genetics could come into play, but not at

present

Instead, the birds seem to have tested their songs,

found out which ones appeared to suffer from auditory masking

And selectively began to use the ‘good’ ones

They had to make the decision based on the behavior of the receivers….and

rank their songs based on these behaviors

So, in sum, we can look at how birds learn their songs and how to use their songs

as a means of examining their cognitive capacities..

and to appreciate the size of a brain no bigger than a pea